CN1360796A - Transmission of compressed information with real time requirement in packet oriented information networks - Google Patents

Abstract

The present invention relates to packet-distributed data transmission of compressed data. According to the invention, parity bits are supplied to the compressed data. The parity bits are used in the entire transmission chain between an encoder having compressed the data, and a decoder which decompresses it. According to one embodiment, the data is speech and the packet-distributed network is a mobile radio network with packet-distribution in included links. However, sending in the radio link of the compressed speech is circuit switched.

Description

Transmission of compressed information with real-time requirements in packet-oriented information networks

technical field

The present invention relates to electronic information transfer, and more particularly to the transfer of compressed data with real-time requirements over at least one packet-oriented transmission link.

technical background

In a fixed telephone network, voice between two users is usually carried over a PCM link. For a voice connection, the PCM link gives a transmission capacity of 64 kbit/s in each direction. Speech in the form of sound is picked up by the phone's microphone, which constitutes an analog speech signal. The analog speech signal can be converted into a PCM coded speech signal by sampling the analog speech signal at a rate of 8k Hz/s, and each sample is quantized and given a binary representation. At this point, the voice is converted into a 64kbit/s bit stream.

In mobile radio networks, speech is sent to the other party via a radio connection between a mobile station and a fixed base station in the mobile radio network.

The number of possible radio connections is limited by the size of the radio spectrum (bandwidth) available to the mobile radio network. Since the available bandwidth is always narrow, it must be used most efficiently. Therefore, in most mobile radio networks, the transfer capacity of a radio link is considerably less than 64 kbit/s. For example, the GSM network has a transmission capacity of 13 kbit/s over voice radio links.

In order to be able to transmit speech with high quality over such a low bandwidth channel, speech is compressed. This is done in a vocoder which encodes an incoming uncompressed speech stream (eg PCM encoded speech stream). The corresponding speech stream output from the vocoder has been compressed so as to have a much lower bandwidth than the incoming signal. At the receiving mobile station, the compressed speech is decoded, after which the original analog signal is reproduced as an audio signal.

Using speech coders, it is also possible to establish a speech connection between a fixed telephone and a mobile station and obtain a high voice quality.

The well known GSM system comprises a plurality of speech coder units. They are centrally located in fixed parts of the network. They are usually located in a base station controller, although they may alternatively be located connected to a mobile switching center. This vocoder unit has a connection to the PCM link, ie a link to the fixed telephone network and the fixed telephone. The speech encoder unit also has a connection to the second link for a transmission capacity of 13 kbit/s speech. The second link leads to a base station with a wireless connection to the mobile station. This second link also has a duplex connection.

The speech encoder unit receives the digitized 64 kbit/s speech stream from the PCM link. Incoming speech is compressed in a vocoder unit. At speech encoding, ie at compression, a number of speech parameters representing the original speech signal are created. One of these speech parameters represents the mouth and how the tone in the speech is formed, while the other represents the harmony. Periods of 20 milliseconds into the stream are encoded and formatted in a speech block. The speech block contains speech parameters and is sent over the second link. A stream of speech blocks sent at 20 millisecond intervals over the second link is received by the mobile station over the wireless link. These blocks of speech are then decoded in the mobile station and the speech is reproduced as speech to the user of the mobile station using the speech parameters.

In the opposite direction, ie speech from the mobile station to the telephone in the fixed network, the analog captured speech signal is sampled and quantized, at which point the speech signal is represented in digital form. Subsequently, the digital signal is speech coded, thereby obtaining speech blocks. This is done in the mobile station. All mobile stations are equipped with a speech coder and speech decoder to encode and decode speech blocks.

The speech blocks created in the mobile station are sent to the speech encoder unit via the wireless link and the second link. In the speech coder unit the speech blocks are decoded and then sent in PCM encoded format over the PCM link to the telephone in the fixed network.

Each duplex circuit switched connection is handled by a corresponding vocoder unit. In the base station controller a plurality of vocoder units is provided to handle a corresponding plurality of circuit switched connections.

A transmission connection shared by the speech coders is provided from the base station controller to the base station. The voice connection is shared by different connection links on a time slot basis. The above-mentioned second link is one of these connecting links. These connecting links have a standardized interface called Abis and each have a capacity for transmitting speech blocks of 13 kbit/s.

For transmission over the Abis interface, a speech block is formatted in a speech frame. In addition to speech blocks, speech frames also contain some control bits so that the speech frames can be received correctly at the base station. The creation of speech frames is done in parallel with the creation of speech blocks in the speech coder. Likewise, the speech frame starts to be sent before the entire speech block is completed in the vocoder. The information in a speech frame is divided into four partial periods of 5 milliseconds each. In speech coding, speech parameters valid for the entire period and speech parameters valid for a partial period are created. When sending speech frames from the vocoder unit over the second link, the speech parameters are grouped together for the respective partial periods, such that the speech parameters valid for the first partial period are sent first, and for the last partial period Valid voice parameters are sent last.

Speech frames are received by the base station and speech parameters are read. When the entire speech frame has been received in the base station, the order in the speech blocks is rearranged. Described in a simplified manner, this is done by grouping together the speech parameters classified as most important regardless of which partial period they come from. These speech parameters are then coded for error detection.

Furthermore, before the speech frame is sent over the wireless link, the most significant bits in the speech frame are channel-coded in the form of convolutional coding with tail bits. Less significant bits remain uncoded. Afterwards, the speech frame is interleaved.

The importance of the speech parameters in speech frames is evaluated by listening trials. This is done in such a way that the listener has the opportunity to evaluate the quality of the decoded speech after an error has been inserted into one of the speech parameters. Errors in some speech parameters appear to cause more serious quality disturbances than errors in other speech parameters. Each speech parameter is represented by a number of bits. These bits have different weights, that is, a bit corresponding to a value of ²² is more important than a bit corresponding to a value of ²⁰ . The importance of all bits included in a speech frame is classified. The classification is based on the importance of the corresponding speech parameter combined with the weight of each bit. GSM 05.03 Version 6.1.2 Table 2 specifies the classification of bits in speech blocks created according to full-rate coding.

Wireless connections between mobile stations and base stations are prone to interference. As a result, some data is corrupted during transfer. Errors in the most important speech parameters can be found and corrected to some extent with the help of error detection coding. Errors in less important speech parameters go undetected.

Technology for speech coding is being developed. As a result, the vocoder units manufactured today are much better than those available when the GSM system was first created. The first vocoder unit that was standardized and used in the GSM system was called "full rate". Two other vocoders were later created. One of these uses only half the bit rate of a full-rate encoder over the air interface, hence the name half-rate encoder. The other uses the same bit rate as the full-rate encoder over the wireless interface, but produces better speech quality, hence the name enhanced full-rate encoder. These three types of encoders are used simultaneously in the GSM system. Different types of vocoders are built into different mobile stations. Therefore, a vocoder unit must be able to handle all types of vocoders.

A base station reorganizes the speech parameters and provides error detection coding for the most important speech parameters in a different way for different types of speech coders.

In patent application WO 97/37466 it is proposed that instead of a standardized interface between nodes in the fixed part of a mobile radio network, packet based transmissions be used. In WO 97/37466, ATM (Asynchronous Transfer Mode) networks are proposed to transfer packet data between nodes. One of the problems that WO 97/37466 attempts to solve is that ATM cells (i.e. packets that carry data over an ATM network) are difficult to fit into the size of a speech frame.

One advantage of using packet based transmission as described in WO 97/37466 is that statistical multiplexing can be used for more efficient transmission over participating links.

To make the transmission more efficient, packet based transmission is used in combination with eg VAD (Voice Activity Detection) and DTX (Discontinuous Transmission). VAD and DTX mean that no content is sent over a voice connection during voice pauses. By using packet-based transmission on the link between the base station controller and the base stations of the GSM network, instead of the standardized Abis interface, and by using DTX and VAD, the link can be used by many more connections than before. The concept used for this increased efficiency is statistical multiplexing. This means that a certain transmission capacity can be used in a more efficient manner since the capacity can be shared between different users in a flexible manner.

There are several protocols for packet-oriented transport. One protocol that is becoming commonly used is IP (Internet Protocol). From IP, an IP packet is created for transmitting the message. The IP packet is provided with an IP header containing information for the connection between the two endpoints.

Another protocol is used on top of IP, such as TCP (Transmission Control Protocol) or UDP (User Datagram Protocol). UDP is mainly used for real-time business. UDP handles the connection between two endpoint applications. UDP creates a UDP message with a UDP header. The UDP header specifies the gateway number at the endpoint, which corresponds to a certain application.

For transport between transport nodes, the functionality is divided into layers supported by a corresponding number of protocols. The relationship between protocols is usually indicated by a protocol stack. Beneath the IP protocol there are protocol layers for different alternatives such as PPP (Point-to-Point Protocol) using HDLC (High-level Data Link Control), ATM or Frame Relay. Layers below IP create packets with accompanying headers, eg derived from HDLC.

A real-time service refers to a service in which users participate in the service interactively and in which transmission delay must be negligible by the user. A telephone call is a typical real-time business, and a video conference with sound and image is another real-time business.

Packet-oriented transmission and IP are basically constructed for the transmission of traditional data services. This kind of data service usually tolerates delay, but is sensitive to data error detection.

Due to the increasing interest in using packet-oriented transport for real-time traffic, work is ongoing to standardize QoS (Quality of Service). This includes giving different types of priority to different types of traffic in packet-oriented transmission. For example, real-time traffic gets high priority for delay and low priority for loss, while the opposite is true for data traffic. Briefly, this is achieved by giving each packet a label specifying according to which priority it should be handled in transit.

Many difficulties will be encountered in meeting the QoS requirements. When an error is detected in a packet, the packet is typically discarded. In the header of a message in the HDLC layer, as well as in the UDP header, a checksum is included to enable detection of errors. If an error is found, the entire packet is discarded. Currently, it is not possible to know where in the packet is located. Although it is possible to ignore error detection in the HDLC layer and it is possible to turn off the function used for error detection in UDP, there is a risk that in the connection to it, there is an important error (such as address in a protocol layer other than HDLC error in ) was not found. This in turn leads to greater errors in the communication (not only in the channel belonging to the connection, but also in other channels).

In the UDP header, besides the gateway address, there is also a checksum. When an endpoint receives a message, the UDP layer compares the contents of the message with a checksum. If the content has been changed, it will be found in the comparison. The location of the error in the message is still unknown.

IETF (Internet Engineering Task Force) is an organization that standardizes the Internet. In a proposal to the IETF, (Larzon, Degermark and Pink) suggested a modification of UDP. This modification is called UDP Lite, meaning that the checksum in the UDP header only covers the UDP header, or alternatively, covers the UDP header with the limited portion of user data that the UDP message is transmitting. Thus in error detection it is possible to know in which part of the message the error resides. But if the checksum works for both UDP headers and user data, it is still possible not to know whether the error resides in the UDP header or in the user data.

Contents of the invention

In a certain node of the transmission chain, data that is compressed in one encoder and sent in real-time through a packet-oriented transmission chain to a decoder where it is decompressed is constantly erroneously detected. The problem solved by the present invention is to be able to reproduce compressed data well in a decoder even if the compressed data is erroneously detected at each node in the transmission chain.

It is an object of the invention to simplify the handling of compressed data in the nodes present in the transmission chain.

In short, the invention proposes that even if the compressed data is distorted during transmission from the encoder to the decoder, it can be sent to the decoder anyway. The decoder decides how to correct for distortion.

The above-mentioned problems are solved by a method of the present invention, in which parity bits are provided to the compressed data in the encoder and sent together with the encoded data through the entire transmission chain to the decoder. In the decoder, the parity bits are compared to the encoded data to detect any errors. The data is decompressed in the decoder, and if errors are found, any errors are hidden during encoding.

The above-mentioned problem is also solved by an encoder unit that compresses a data stream by creating parameters representing data in the data stream. These parameters are divided into data blocks, and the positions of the parameters in the data blocks are ordered according to the importance of the parameters relative to each other. Data blocks are provided with parity bits so that errors in transmission can be found. As an alternative to ordering the positions of the parameters in the data block, the positions of the bits representing the parameters are ordered according to their importance.

Decoders such as mobile stations have hitherto been equipped with good means for concealing detected errors when decompressing received compressed data. However, errors found in the mobile station are only errors caused in the radio link to the mobile station. An advantage of the invention is that the means used by the decoder for concealing errors can also be used for errors occurring in other transmission links than the radio link.

Yet another advantage of the present invention is that nodes considering compressed data transmission do not need to know how the data is compressed in order to process the data correctly. The introduction of new types of speech coding is thus quite simple. For example, the vocoder unit needs to be changed only when a new type of vocoder is inserted in the mobile station. Speech coder units are present in fewer nodes than in base stations, therefore, they are easier to upgrade.

The invention will now be described in more detail with the aid of preferred embodiments and with reference to the accompanying drawings.

Description of drawings

Figures 1a and 1b show block diagrams of connections in previously known connections for establishing a voice connection between a mobile station and a telephone in a fixed telephone network.

Figure 2 represents a diagram of the format for a previously known speech frame formatted for the Abis interface.

Figure 3a shows a previously known graph of speech parameters ordered by importance in a speech frame.

Fig. 3b shows a previously known graph of the same speech parameters as shown in Fig. 3a after error detection coding has been applied to some speech parameters.

Figure 3c represents a previously known graph of the same speech parameters as shown in Figure 3b after further reorganizing the speech parameters and providing additional tail bits.

Fig. 3d shows the previously known format for the speech parameters in Fig. 3c after a convolutional code has been provided to some of the speech parameters.

Figure 4 schematically represents a previously known protocol stack.

Figure 5 shows a block diagram of a possible structure for a third generation mobile radio network and transport network.

preferred embodiment

The first technical condition required for understanding the present invention will now be described.

Figure 1a shows a mobile station MS in a mobile radio network PLMN. The mobile radio network PLMN in FIG. 1a is a GSM network. A voice connection is established between the mobile station MS and a telephone in the fixed public telephone network PSTN. The call is connected to the fixed telephone network PSTN via a base station BTS and a speech coder unit 11 (transcoder and rate adaptation unit) in the GSM network 11 . Although there are many more nodes participating in the GSM network PLMN, Fig. 1a only shows those which are of interest to the invention. However, the nodes and interfaces shown in Figure 1a are previously known. The speech coder 11 is implemented in the base station controller.

Between the telephone TLP and the speech coder unit 11 of the fixed network PSTN, the call is carried over a PCM link. The PCM link carries calls in the form of quantized samples generated with a sampling frequency of 8K Hz. This results in a 64kbit/s voice stream delivered to the PCM link. 64 kbit/s is the capacity normally used for connecting calls between telephone TLPs in the fixed network PSTN.

The call is transferred between the mobile station MS and the base station BTS via a radio connection RL, between the base station BTS and the speech coder unit 11 via a fixed link A (hereinafter referred to as Abis link A) with an Abis interface. The radio link RL and the Abis link A have a transmission capacity for speech of 13 kbit/s.

In the speech encoder unit 11 the incoming speech stream of 64 kbit/s from the telephony TLP is compressed. The corresponding voice stream is obtained from the vocoder, but at 13 kbit/s. The compressed speech stream is sent to the mobile station MS over the Abis link A and the radio link RL. In the mobile station MS speech is decoded and converted into sound.

The speech captured as sound in the mobile station MS and to be transmitted to the telephone TLP is recorded by the microphone, thereby forming an analog signal. The analog voice signal is sampled and quantized to obtain a digital voice stream. This digital speech stream is coded in the mobile station MS in the same way as in the speech coder unit 11 and sent to the speech coder unit 11 via the radio link RL and the Abis link 12 . The speech is decoded in the speech encoder unit and sent over the PCM link 13 to the telephone TLP as a 64 kbit/s sampled speech stream.

The incoming speech stream from the PCM link to the speech encoder unit 11 is divided into periods of 20 milliseconds. For each 20 millisecond period, the speech encoder unit 11 forms a speech block SPB containing speech parameters representing the speech. Figure 3a shows a speech block SPB obtained by speech coding according to "full rate". The speech block SPB contains 260 bits representing speech parameters. Each parameter is represented by at least two bits. These bits correspond to the nominal values 2 ⁰ , 2 ¹ , 2 ² etc., where the bit corresponding to 2 ² has higher weight than the bit representing 2 ⁰ . The importance of different speech parameters is given an objective evaluation by means of listening trials. In the specification GSM 05.05 version 6.1.2 Table 2, the importance of each bit in a speech frame is classified. The classification is based on the importance of the corresponding speech parameters and the weight of the bits. These bits are divided into class I and class II, where class I means a bit is more important than a bit of class II. Within class I there are two groups, Ia and Ib, with Ia being more important than Ib. Group Ia includes 50 bits, group Ib includes 135 bits, and class II includes 78 bits according to full rate coding.

In Fig. 3a, the positions of the bits in the speech block SPB are ordered according to importance, group Ia first in the speech block SPB, followed by bits of group Ib and finally bits belonging to group/class II. However, in the speech encoder unit 11 the bits in the speech block SPB are not sorted according to their importance, but are sorted later when the speech block SPB has been sent to the base station BTS.

Before the speech block SBP is sent to the base station BTS, it is formatted and given control/check bits, thus forming a speech frame SPF. The voice frame SPF is shown in Figure 2. A Speech Frame SPF consists of 20 double octets, each corresponding to a row/row in the Speech Frame SPF. Each double octet contains 16 bits. The first double octet in a speech frame consists of zeros and the first bit of the remaining double octets consists of ones. The following two octets CNTB contain the check digit. This is also described in GSM 08.60 Version 5.1.1 (February 1998). The following double octets in the speech frame SPF are referred to as five groups 21-25, as indicated by dashed lines in FIG. 2 .

The first two-octet group 21 is used to represent the speech parameter group valid for the entire 20 millisecond period. The four following groups of double octets 22-25 are used to indicate the speech parameters valid for the corresponding 5 ms partial period out of the total 20 ms period. Thus, the second group 22 contains the speech parameters valid for the first 5 ms sub-period out of the total 20 ms, and the following group 23 represents the speech parameters valid for the second 5 ms sub-period etc.

The data of the speech block SPB corresponds to a bit rate of 13 kbit/s via the Abis link A. However, the data of the speech frame SPF corresponds to a higher bit rate of 16 kbit/s via Abis link A.

The creation of the speech frame SPF is simultaneous with the creation of the speech block SBP, so the speech parameters need to be sorted in time order. Start sending the speech frame SPF before the entire speech frame SPF is complete. The reason for this is to use the 13 kbit/s transmission capacity of the Abis link as efficiently as possible to avoid delays.

When the base station BTS receives the entire speech frame SPF, the speech parameters are read out of the speech frame SPF and the positions of the bits in the speech block SPB are sorted according to their importance. The sequence shown in Figure 3a is thus obtained. In order to be able to detect any situation in which speech parameters belonging to group 1a are received erroneously at the mobile station MS after the radio transmission, three parity bits CRC are provided to the bits in the speech frame SPF corresponding to the speech parameters in group 1a , see Figure 3b. The parity bit CRC is provided according to the error detection principle of the cyclic coding "Cyclic Redundancy Check". This cyclic code is a block code.

Afterwards, the speech parameters according to class I are again reordered and provided with four tail bits TAIL, see Fig. 3c. Speech parameters in class I are convolutionally coded, thereby increasing the number of bits from 189 to 378 bits. A convolutional code makes it possible to correct a limited number of errors in the class I bits after reception in the mobile station MS. Certain positions are given better protection than others by the convolutional code, and in the reordering, the most important speech parameters are placed in the most protected positions. After the 378 class I bits, 78 bits representing class II speech parameters are placed without error detection coding. Figure 3d shows a coded speech block CSPB, which now comprises 456 bits. The coding of the speech block CSPB in Fig. 3d is thus performed in the base station BTS.

Before sending the coded speech block CSPB over the radio link RL, the speech block is interleaved, ie divided to be sent in a number of TDMA bursts to the mobile station MS.

The previously known functions in the GSM network PLMN are described above to facilitate the understanding of the invention. A prerequisite of the invention is the use of packet-oriented transmission over the fixed link between the speech coder unit 11 and the base station BTS instead of circuit-oriented transmission. The Abis interface implies circuit-oriented transport. The link previously referred to as Abis link A with reference to FIG. 1 a is hereafter referred to as fixed link 12 . Next, Figure 1b is discussed. The difference between Figures 1a and 1b is that on the same link, in Figure 1a circuit-oriented transmission is used between the speech coder 11 and the base station BTS and in Figure 1b packet-oriented transmission is used.

According to the invention, error detection coding is provided for speech blocks SPB in the speech encoder unit 11 instead of the base station BTS. Error detection coding is then used during transmission over the fixed link 12 and over the radio link RL. Thus, when a speech block is received by the mobile station MS, errors can be detected regardless of whether they occurred during transmission over the fixed link 12 or over the radio link RL. In the uplink, the mobile station MS provides a created speech block SPB with error detection coding and sends it over the radio link RL. According to the invention, the speech blocks SPB are sent to the speech encoder unit 11 via the fixed link 12, leaving the error detection coding.

For speech coding according to "full rate", the speech encoder unit 11 orders the speech parameters according to their importance in their position in the speech blocks SPB, so as to obtain an order as shown in Fig. 3a. The 50 bits in class Ia are then provided with three parity bits according to a "cyclic redundancy check". With the three parity bits CRC forming the code for error detection, the speech block SPB contains 263 bits.

A coded speech block SPB having 263 bits is provided with parity bits, constituting a speech frame SPF. Speech frames SPF are sent to the base station BTS. The speech frame SPF does not have the same format as the speech frame sent on Abis link A, but, in principle, it has the same format as the speech block in Fig. 3c, provided with an extra parity bit. The parity bit specifies the type of coding used, etc., so that the decoders in the base station BTS and the mobile station can correctly process the speech block SPB. In the base station, a speech block SPB with 263 bits is read and the class I bits are reordered to provide error correcting convolutional coding in the manner described above in connection with Figures 3c and 3d. At this time, the speech block SPB includes 456 bits. These 456 bits are interleaved and sent to the mobile station MS in multiple TDMA bursts.

When using packet-based transmission, the speech coder 11 creates the entire speech block SPB for a period of 20 milliseconds before the speech block SPB is sent over the fixed link 12 . This makes it possible for the speech encoder units to order speech parameters according to their position in speech blocks SPB.

The above-described embodiment shows how to encode a speech block created according to "full rate" speech coding. Speech blocks SPB created from "half-rate" and "enhanced full-rate" speech coding appear in a similar manner. This means that the positions of the speech parameters in the speech block SPB are ordered according to their importance. Afterwards, important speech parameters are provided with a parity CRC according to a "cyclic redundancy check". This occurs in the speech coder 11 in the same way as it occurs in the current base station BTS.

For "enhanced full rate", before the speech block SPB is sent from the speech encoder unit 11, important speech parameters in the speech block SPB are already provided with parity bits according to the "cyclic redundancy check". However, the speech parameters are not ranked according to their importance before encoding, the purpose is only for error detection during transmission over the fixed link 12 . When a speech block SPB has been received in the base station BTS, the parity bits provided by the speech encoder unit 11 are cleared. Instead, the base station BTS reorders the position of the speech parameters in the speech block SPB and then provides the parity bit CRC to the important speech parameters. The parity bits CRC provided by the base station BTS are used for error detection during transmission over the radio link RL.

Fig. 4 shows a protocol stack PSCK with different protocol layers used during the transmission between the speech coder unit 11 and the base station BTS. Applications reside at the highest level of the stack PSCK. In this case the application is speech coded speech in the form of speech blocks SPB with control information for decompressing the speech. Speech-encoded speech is transferred between the mobile station MS and the speech encoder unit 11 by the application layer.

The underlying UDP layer is used to transport the application.

Below the UDP layer there is an IP layer. The IP layer handles the transport traffic between two endpoints of a transport chain (usually between a server and a router or between routers). Since the standardized radio link RL in GSM is not suitable for packet transmission, the end point for IP is formed by the speech coder unit 11 and the base station BTS. In Fig. 1b, the router between the speech coder unit 11 and the base station BTS is not shown, but there should be a router for traffic control in the realized network.

Below the IP layer there is a layer PPP (Point-to-Point Protocol), which is carried by HDLC (High-level Data Link Control), below which is for example a layer SDH (Synchronous Digital Hierarchy) or PDH (Pseudo-Synchronous Digital Hierarchy), where E1 (2048M bits/second) is usually provided in the mobile telephone network. For the layers below IP, an alternative protocol to the protocol shown in Figure 4 can be used, eg ATM (Asynchronous Transfer Mode).

One speech block SPB transmitted from speech encoder unit 11 in downlink is carried by one UDP message. The UDP message contains speech blocks SPB and a UDP header. The UDP header contains the gateway number to a receiving application in the base station BTS, ie a logical traffic channel using a specific time slot in the radio interface, and a sending gateway in the speech coder unit 11 . The UDP header also contains a checksum (i.e., a parity bit) to detect any instances of distorted data during transmission.

The IP layer encapsulates UDP messages in an IP packet. The IP packet comprises, in addition to the UDP message, an IP header with an IP address to said base station BTS. Using the IP address and the UDP gateway number, speech blocks SPB are identified which belong to the speech connection between the mobile station MS and the fixed telephone TLP. The IP header also contains a checksum, but this only applies to the IP header.

The HDLC layer, ie the layer that carries the IP packets in the HDLC packets, also provides the HDLC packets with a checksum that works over the entire HDLC packet.

Packet-oriented transmission and the protocol stack PSCK according to FIG. 4 are generally used for data transmission. Unlike voice, data is sensitive to error detection but can handle delays. When an error is found with the help of a checksum in an HDLC packet, retransmission of the data is usually requested. Requests for short delays make retransmission of a speech block impossible if the checksum indicates an error.

The checksums in both the HDLC header and the UDP header can be set to zero, where no error correction is performed to check that the UDP message was detected correctly. According to the invention, both checksums are set to zero, so that even if an error occurs in them during transmission from the speech encoder unit 11, the speech block SPB can be transmitted to the mobile station MS.

As an alternative to setting the checksum in the UDP layer to zero, use a checksum which, according to the proposal UDP Lite, works only on the UDP header or only on a limited part of the UDP header and content.

The speech coder in the mobile station MS and the speech coder unit 11 are equipped with the function of effectively concealing errors in the speech block SPB when the speech block SPB is decompressed. This functionality is already plugged into existing mobile stations MS.

With previously known techniques, it is possible to use an efficient method for the mobile station MS to process erroneously detected speech blocks SPB only for errors occurring during transmission over the radio link RL. Thanks to the invention, errors occurring during transmission over the fixed link 12 can also be handled in the speech coder of the mobile station MS.

Without the present invention, and setting the checksums in HDLC packets and UDP headers to zero, it would not be possible to find errors in speech blocks SPB occurring in the fixed link 12 . On the contrary, if a checksum is used, the entire speech block SPB will be discarded when an error detection occurs, even if the error only concerns less important speech parameters.

The speech coder unit 11 is included in currently available mobile radio networks as shown in the GSM network PLMN. In third generation systems, it is proposed that the speech coder unit 11 will not be part of the fixed part of the mobile radio network. Instead, the speech coder unit is present outside the mobile radio network in so-called media gateways.

Figure 5 shows a possible structure for a third generation system, 3G, in which telecommunications and data communications are mixed. Currently, a functional division is given instead of dividing the network into a data or telecommunication network. The transmission network 51 (backbone network) only maintains transmission services. Services as conventional telephony as well as Internet communication are provided by separate access networks 52, 54, 55 using the transport network 51. The access networks 52 , 54 , and 55 access the transmission network through the media gateway 53 . One of the access networks 52 , 54 , 55 is the radio access network 52 . The radio access network 52 provides bearer services for communication with mobile stations MS over radio links RL. The radio access network 52 includes a plurality of radio base stations BTS. The radio base station BTS is applicable to different radio access technologies.

Several functions inserted in the mobile radio network PLMN are currently being dispensed with in the radio access network 52 . Rather, these functions are provided as separate services available via the transport network 51 . One of such functions is eg mobility (mobility management), which makes it possible to route an incoming call to a certain mobile station MS independently of where the mobile station MS is located in the coverage area of the radio access network 52 .

Another function not included in the radio access network 52 is speech coding. A mobile station MS using the radio access network 52 for communication is provided with a speech coder. Speech coders in mobile stations can be of various types. Speech coder units 11 corresponding to those plugged into the GSM network are present in the media gateway 53 . When voice communication is performed between a mobile station MS connected to the telephone access network 54 and a telephone TLP, a coded compressed voice stream passes through the radio access network 52 and the media gateway from the mobile station MS through the radio link RL 53, and send it to the media gateway 53 connected to the telephone access network 54 through the transmission network 51. In the following media gateway, the voice stream is decoded. The speech stream is then sent decoded via the telephone access network PSTN at a higher bit rate, for example as a PM-coded signal.

In the opposite direction, ie from the telephone TLP to the mobile station MS, the voice stream is sent uncompressed via the telephone access network PSTN to a media gateway 53 connected to the transport network 51 . In the media gateway 53, the voice stream is compressed by encoding. Then, the compressed stream is sent to the mobile station MS via the transport network 51 via the media gateway 53 via the wireless access network 52 . In the mobile station, the voice stream is decoded depending on which sound is created.

In the transmission between the mobile station MS and the media gateway connected between the transport network 51 and the telephone access network 54, packet-oriented transmission is used. It is possible to use a UDP layer and an IP layer as shown in Figure 4, but alternative protocols could also be used.

During speech coding, the type of speech coder used in the mobile station MS as well as in the media gateway 43 divides the speech into short periods and creates speech parameters for each period. The speech parameters are then sent in speech blocks in a manner similar to speech coding according to full rate, half rate and enhanced full rate.

Another advantage is that many different types of vocoders can be used in the media gateway 43 without the nodes involved in transmitting speech blocks SPB having to adapt the transmission to the type of vocoder used.

The invention is more important for the third generation system 3G than for the current mobile radio network PLMN since the encoded speech will be sent through more nodes, thereby increasing the risk of the data being misinterpreted. If an error is found, it is important that the error is handled by the vocoder and not by one of the nodes handling the transmission.

In addition, not only voice but other real-time services are also compressed and sent. Some of these services, such as video services, require large bandwidth. In order to use the transmission network 51 efficiently, the information stream is compressed, for example by video coding.

For example, a duplex video connection can be established in real time between the first terminal T1 connected to the wireless access network 52 and the second terminal T2 connected to the Internet 55 . Terminals T1 and T2 are shown in FIG. 5 . Each terminal T1 and T2 is equipped with a video and speech encoder and decoder. Video and voice signals are sent between two terminals T1 and T2. According to the invention, compressed speech and video signals are provided with error detection coding.

It should be noted that the invention is important even if no radio link RL is included in the transmission chain used to compress the information.

Coding may be provided for selected speech parameters or bits even if the positions of the coded bits or speech parameters are not ordered in order of importance. However, if the positions of bits and/or parameters are ordered, this should be done in the encoder. The nodes involved in the transmission chain thus do not need to be further reordered and the transmission is simplified.

In addition, there is another important reason for ordering the position of the most significant bits or speech parameters in the speech block SPB. During packet-oriented transmission over fixed links 12 and in third-generation networks 3G, overloading sometimes occurs. Delays occur and packets are dropped as a result of the overload.

Sorting the position of the speech parameters or bits in the speech block SPB simplifies the ability to discard less important speech parameters or bits during overload on the link 12 or any link of the radio access network 52 or transport network 51, so that The more important speech parameters or bits are sent over link 12 . The present invention does not give a complete solution for discarding less important speech parameters/bits during congestion, but ordering the speech parameters/bits is an important prerequisite for this to be possible. The present invention is advantageously used in conjunction with the invention described in US Patent Application Serial No. 09/275069. The latter patent application gives a solution how to drop parts of packets during overload.

By sorting the position of the bits according to their importance in the speech block SPB, it is also possible to divide the speech block into different speech frames SPF and have the speech frames sent in packets with different priorities. For example, more important bits are sent in a packet given high priority to reach the endpoint, while less important bits are sent in another packet with lower priority to reach the endpoint. The criteria for QoS make it possible to send packets with different priorities.

"Circuit-oriented" means that of the total transmission capacity in the link 12, there is a certain capacity per time unit for each of a plurality of different connections. It is also assumed that the link 12 handles the transmission of a plurality of speech connections between the base station BTS and a corresponding number of speech coders 11 .

Here, "packet oriented" means that the total transmission capacity in the link 12 is shared between all ongoing speech connections, and the capacity is allocated to the one connection that has information to send at the moment. Speech is transmitted using packets of a pre-defined format. DTX (Discontinuous Transmission) is a technique for detecting pauses in speech and for terminating the creation of speech blocks SPB during the pauses. Thus, with DTX, fewer speech blocks SBP are sent over the link 12 and, for each speech connection, less transmission capacity over the link 12 is required when viewed over a longer period of time.

Here, "error detection coding" refers to providing additional bits, such as parity bits, to the information being transmitted so that when the information is received, it may be compared with the extra bits. There are several types of error detection coding. The most common group is block encodings. A "cyclic redundancy check" is part of chunk coding.

Of course, the invention is not limited to the embodiments described above and shown in the drawings; rather the invention can be modified within the scope of the following patent claims.

Claims (16)

1. one kind is used for the method that real-time Transmission has the digitalized data stream of first bit rate, comprises step:

-at first node (11, MS, 53) by described data stream encoding is compressed,

Thereby obtain to be significantly less than second bit rate of first bit rate,

-send packed data stream by connection (12,52,51) towards grouping,

-at Section Point (11, MS, 53) described data flow is decompressed, thus recover first

Bit rate,

Its feature also is step:

-after the first node compression, provide parity check bit (CRC) to give described data flow, from

And described data flow obtains three bit rate slightly higher than second bit rate,

-at Section Point described parity check bit (CRC) is compared with described data flow, with

Find in the data flow by the data of error detection.

2. the method for claim 1, wherein first with second node in one be the travelling carriage that is connected (MS) that has by a wireless links (RL).

3. method as claimed in claim 1 or 2, wherein when data flow was compressed, this data flow was divided into corresponding to the section of determining the length time cycle, and was data block (SPB) that contains the parameter of the data of representing this section of each section establishment.

4. method as claimed in claim 3, wherein the importance of parameter relatively is classified mutually, and according to the position ordering of importance to parameter in the data block.

5. method as claimed in claim 4 wherein according to its importance parameter is divided into two classes, and the parameter in the most important class is provided for the described parity check bit (CRC) of error checking.

6. method as claimed in claim 3, wherein in data block, each parameter is represented by at least two bits with different power, and according to described power is sorted in the position of two bits in the data block.

7. method as claimed in claim 6, the bit that wherein has Gao Quan is provided the described parity check bit (CRC) that is used for error checking.

8. method as claimed in claim 3, the wherein speech of data flow composition digital translation, and data block (SPB) is a block of speech (SPB), parameter is a speech parameter.

9. method as claimed in claim 3, wherein data flow is the vision signal of digital translation.

10. method as claimed in claim 3, though wherein when sending data block (SPB) arrived by error detection, data block (SPB) also is sent to second node (MS, 11,53).

A 11. cell encoder (11), it has the device of the data flow that is used to receive first bit rate, also has device, be used for coming packed data stream by data flow being divided into corresponding to the section of partial periodicity, and for each partial periodicity is created a data block (SPB) that contains the parameter of the data in the expression correspondent section, thereby produce data block (SPB) stream with second bit rate that is significantly less than first bit rate

It is characterized in that:

Device is used at each data block (SPB), for as the parameter of this data block part or alternative bit, and according to the mutual importance of described parameter or alternative described bit rank order according to prior regulation, and

Device is used to provide parity check bit to data block (SPB), with the mistake of finding to occur in data block (SPB) transmission.

12., have and be used for when entering data flow device to this data flow speech coding when representing voice according to the cell encoder of claim 11.

13., have and be used for when entering data flow device to this data flow video coding when forming a vision signal according to the cell encoder of claim 11.

14. cell encoder according to claim 11, has the device that is used to receive the data block with the 3rd bit rate (SPB) stream that contains parameter, also has device, be used for the parity check bit (CRC) that identification has offered data block (SPB), with the bit in the data block and described parity check bit comparison to find mistake, also have device, be used for that the parameter decoding is had the data flow of four bit rate higher than the 3rd bit rate thereby create.

15. mobile wireless network (PLMN) that has according to any one cell encoder of claim 11-14.

16. a mobile wireless network (PLMN) comprising:

At least one is speech coder unit (11) fixedly; It has one to the connection of duplexing PCM link, one to the connection towards the link (12) of grouping; Have for compression from the voice flow of PCM link and with its device that transmits by the link (12) towards grouping as block of speech (SPB) stream with compressed format; And have for receiving the device that block of speech flows from the link (12) towards grouping; And be used for the block of speech decoding and form the device of the voice flow that pass through the transmission of PCM link of a decompression

Be connected to towards the link of grouping and be connected at least one base station (BTS) of at least one Radio Link (RL), it has the device that is used for from receiving block of speech (SPB) stream towards the link that divides into groups and block of speech stream being passed through Radio Link (RL) transmission, also have and be used for the device that receives block of speech stream and they are transmitted by the link (12) towards grouping from Radio Link (RL), and

A travelling carriage (MS), it has the device that is used for receiving from Radio Link (RL) block of speech (SPB) stream, be used for device to the voice flow of block of speech (SPB) decoding formation decompression, the device that is used for electrographic recording sound, be used to compress the device that has write down voice, wherein block of speech (SPB) is formed, and the device that is used for sending by Radio Link block of speech (SPB), it is characterized in that:

Be used for providing the device of parity check bit for the block of speech (SPB) that is created at speech coder unit (12) and travelling carriage (MS), and

Device, the content that is used in travelling carriage (MS) and speech coder unit (11) block of speech (SPB) that will be received and subsidiary parity check bit comparison are to find possible mistake, occur when wrong with box lunch, can be in the decode procedure of reception block of speech (SPB) concealing errors.

Images